As technology marches forward, enabling food designers to employ increasingly authentic substitutes for macronutrients like fat and sugar, the science of safety testing is being forced to move ahead too.

Nothing has changed as far as the U.S. Food and Drug Administration is concerned. Petitioners seeking approval for food additives and ingredients that might be consumed at the level of a gram per day, such as sugar or fat substitutes, must conduct their toxicology studies in accordance with the same regulations developed for additives that will be consumed at the milligram level, such as preservatives.

But, where FDA normally requires a 100-fold margin of safety in applying animal-study results to humans, it sometimes isn't possible to feed macronutrient additives to animals at levels high enough to achieve such a large margin.

The solution would appear to be obvious: Lower the safety margin in animal testing, and increase the emphasis on human testing. But there's a catch. Although FDA has accepted data from scientifically sound clinical studies in these situations, the agency does not require clinical testing on humans - leaving researchers with no guidelines for their clinical studies.

The response of ingredient suppliers to this dilemma has been to redouble their efforts to ensure that the animal models they're using, albeit at lower margins than 100-fold, are metabolically similar to humans. And thus far, at least some of these petitions have earned FDA approval.

"By scientific agreement, there are times when [clinical testing] would be useful," confirms George Pauli, Ph.D., director of the Division of Product Policy in FDA's Office or Premarket Approval.

Once there has been at least enough animal testing to satisfy the manufacturer that the substance will not harm humans, the next step is to find that metabolically similar animal model. Current regulations call for testing in these specific areas:

To these, FDA is about to add a fifth test:

The object of this series is to glean enough data to be able to select an animal model that's a good match for the human response. Researchers watch not only the animal's response, but also whether different animals respond the same or differently. For example, researchers at McNeil Specialty Products Company found that three different species - rats, mice and dogs - had almost the same response to sucralose, the non-nutritive sweetener that FDA is currently reviewing.

Once the animal data is collected, the next step is to perform the same series of tests on human subjects. "When you do those studies in man, you've got your fingers crossed," says Leslie Goldsmith, V.M.D. and Ph.D., vice president of product safety and regulatory affairs.

If the findings from the human subjects indicate that the animal model is not "relevant to the human condition," Goldsmith adds, then it's back to square one on the search for a new animal model. But if the findings do correlate, they can considerably strengthen the case for approval. This could be especially important with a macronutrient additive that has a safety margin less than 100-fold.

Once the substance progresses to tolerance studies, which are designed to assess the results of exposure, as opposed to how it is metabolized, clinical testing is a given. There are also instances when only human testing will provide the necessary information, especially if it is necessary to determine how the substance will affect a specific segment of the population. Any sweetener that's designed to replace sugar, for example, has certainly been tested on human diabetics volunteering for these tolerance studies - unless animal studies have shown that the product does not affect glucose metabolism.

Even after a substance has been approved, manufacturers could have reason to perform still more clinical trials. Anyone who thinks the job is done once FDA has accepted the petition would do well to recall the additional time and money invested in saccharin, when FDA repealed its approval in the 1970s, and aspartame, when the PKU (phenylketonuria) question arose.

In the case of saccharin, it was shown that although rats did contract bladder cancer, they did so only when fed doses that far exceeded what humans could consume. And in the case of aspartame, The NutraSweet Co. found after further testing that phenylketonurics could be adversely affected and has labeled its products accordingly.

The cost of these clinical studies is one of the main reasons why the largest food or drug companies are almost always the ones to develop truly groundbreaking products.

"A group shouldn't start this process unless they have hundreds of millions of dollars that they can put at risk," Goldsmith says. "You can count on a decade of preliminary research."

The review process itself can take several years; Goldsmith adds that "it would probably behoove you to have a manufacturing facility ready to go."

Even so, the burden is smaller for food companies than for drug manufacturers. The first few rounds of "ADME" tests, which determine whether humans metabolize the substance similarly to the animal model, often involve fewer than six subjects. The numbers increase from there, especially for longer-term tolerance studies, which can involve hundreds of subjects.

There is no exact number of subjects required under FDA regulations. One determining factor is "how rigorous the tests are," according to Sandy Bigelow, Ph.D., associate director of food safety and regulatory affairs at Pfizer Inc. Food Science Research and Development. "The more rigorous the tests, the fewer people, in general," he notes.

Although food companies need to "spend their money wisely," the tests also must "be determinative" - meaning statistically significant, Bigelow adds. "You want to provide FDA with definitive studies," he explains, but "you shouldn't have to provide gobs and gobs of data to establish safety, just sufficient data."

To an industry outsider, it may seem that "sufficient data" based on so few subjects is hardly sufficient to warrant FDA approval. But the research that is conducted actually is quite rigorous. The same double-blind, placebo-controlled methods specified for drug testing also apply to food ingredients. The research is always conducted by an "institute of high regard," says Goldsmith, and it must pass an equally well-regarded Institutional Review Board (IRB).

One key point of departure from drug-testing regulations is that once the petition is submitted, all the research becomes public knowledge. But even that contributes to the accuracy of a company's studies: Not only do the researchers have access to related studies that have already been published, they also have the opportunity to review related studies submitted by their competitors.

Certainly some manufacturers might be less inclined to invest several million dollars to develop data that they will eventually have to share. Ironically though, the fact that so much information is available to researchers increases the odds that their petitions will be approved. And in the long run, it may hone the procedures for both human and animal testing to the point where manufacturers could do less of both without sacrificing public safety.